[#235] Add a custom 404 page

[vharseko]

Summary

The site had no 404.html, so any broken or mistyped URL showed the generic GitHub Pages "404 — File not found" page with no way back into the site.

• 404.html — a new page on layout: home (navbar, footer, cookie notice) with permalink: /404.html. It explains that old blog / GitHub wiki links have most likely moved to the blog, and links to Home, Blog, Documentation, Support, GitHub Discussions and every product page (taken from _data/products.yml, the same source as the Products menu). All links are absolute, so the page works when GitHub Pages serves it under any nested URL.
• sitemap.xml — pages with sitemap: false are skipped; the 404 page uses it.
• _includes/header.html — renders <meta name="robots"> when a page sets robots:; the 404 page sets noindex, since /404.html itself is served with status 200 when requested directly. No other page sets it, so their output is unchanged.

Fixes #235

Verification

Built the site with bundle exec jekyll build (Ruby 3.1, as in CI) and ran the same htmlproofer internal-link check as the workflow: 375 files, no errors.

• _site/404.html: <title>Page Not Found · Open Identity Platform</title>, <meta name="robots" content="noindex">, navbar and footer present, five product links.
• /404 is not in sitemap.xml.
• /, /openam and a blog post have no robots meta tag, as before.

[maximthomas]

praise: The change is small, sits where each behaviour lives, and leaves every other page's output unchanged.

• sitemap.xml:29 — page.sitemap != false is true for every page without the key (Liquid 4 != is !(nil == false)), so only the 404 page drops out of the sitemap.
• _includes/header.html:25-27 — {%- if page.robots %} / {%- endif %} trim exactly the newline they stand in for, so the head of a page without robots: renders as at base.
• 404.html:42-44 — the product links come from site.data.products, the same source as the Products menu, so the page cannot drift from the navbar.